Skip to main content

Full text of "Languages and Dialects of China"

See other formats


2.1 Phonetic Features of Wu dialects. 

Modern Wu dialects characterized by implosives or preglottalized series 
as a reflex of Bang and Duan initials are Qingtian, Qingyuan, Jingning, 
Jinyun, Yongkang, northern Yongjia, southern Xianju, and a part of 
Wencheng in southern Zhejiang and Jinshan, Songjiang, Fengxian, 
Quansha, Xinzhuang and Nanhui in Shanghai region. There is no contrast 
between pb], [id] and [p], [t] in these dialects, and they are in free variation 
in some places. The realization of Bang and Duan initials as implosives is 


therefore only a nondistinctive trait of speech. There is a difference in the 
degree of glottal constriction between dialects in these two regions. Dialects 
in southern Zhejiang are not only characterized by a sharp glottal 
constriction but also by a conspicuous implosive gesture and more 
backward tip of tongue contact. Implosives might have had a wider 
geographical distribution; even today there are still a handful of people in 
the urban area of the city of Shanghai who preserve the implosives in their 
speech (Xu and Tang 1982). Bang and Duan initials in even earlier 
Shanghainese should also have been pronounced implosives (Chao 1928). 
The Wenzhou dialect as recorded in Parker (1884) has true voiced initials b 
and d in upper and lower tone categories. He thinks that the two initials are 
unaspirated in the upper tone categories and should not be confused with 
/p'-/ and /t'-/ whereas aspiration makes no difference in their 
counterparts in the lower tone categories. Hirayama (1983) concludes, on 
the basis of Parker's description, that Bang and Duan initials become voiced 
in the Wenzhou dialect. But a comparison with the realization of Bang and 
Duan initials in the Yongjia dialect will however further clarify the issue. 
b- and d- in upper tone categories in Parker's terms are in fact implosives 
5 fc- and Id-. Since phonetic properties of implosives were not known then 
he took them for b- and d- 9 but he senses that they are different from the 
b- and d- in yang tone categories. So he spills some ink explaining it. 

Although implosives have disappeared from most Wu dialects, the 
traces they left distribute over a wider region. The main traces are 
discussed below: 

(1) The nasalization of Bang and Duan Initials in Words with nasal 
endings: (*b,*d -->*m,*n/ Vn(^)). 

This phenomenon is treated in detail in Hirayama (1983) where its 
phonetic rationale is also explicitly given. In the articulation of implosives 
the lessening of air pressure in the oral cavity may well lower the soft 


palate so that nasal sounds are liable to be induced. For example, little 
airflow escapes through the orifice near the soft palate into the throat after 
the release in the event of producing implosives in the Chuansha dialect "so 
that it strikes the listener as having a faint trace of the nasal sound. That is 
why for Chuansha speakers the nasal sounds in the Pudong dialect are too 
heavy." (Shi 1985). 

The nasalization of Bang i & and Duan t ^/initials is embraced by the 
Wuzhou region and the Liqu region in central Zhejiang. Some individual 
words like 'beat' / c n» / in the Pucheng dialect of northern Min also exhibit 
the same phenomenon. They are mainly of the following types: 

Type 1. Only individual words undergo nasalization, Qingtian 

representing a most extreme case where only 'beat' is 

realized as /*ne /. 
Type 2. Bang and Duan initials are nasalized when finals in 

modern speech are nasal, as in Jinyun. 
Type 3. Bang and Duan initials are nasalized as long as the words 

in question have no medials and have a nasal ending in 

Ancient Chinese, as in Yongkang. 
Type 4. Bang and Duan initials are nasalized as long as the words 

in question have a nasal ending irrespective of whether 

there are medials, as in Yiwu. 

There is a gradual increase of nasalized words and a gradual decrease 
of implosive elements from Qingtian to Yiwu. This cline reflects the 
geographical distribution of the change of implosives to nasals. 

(2) The Realization of the Duan Initial as the Lai Initial. 

In the genetically related languages examples of correspondence 
between fld/ and HI are common, as exemplified by 'good' / 2 dai/ in the 
Liyan variety of the Sui language and /lai/ in the Jinping variety of the 
Kam language and 'name' / daan/ in the Mak language and /laan/ in the 
Yanghuang language. The phonetic explanation has yet to be worked out. It 
might well be related to the fact that there is a longer duration of closure in 


the production of the implosive fid/ and the downward movement of the 
throat causes the tongue to move backward so that the closure of the tip of 
the tongue becomes loosened. The Duan initial is realized as HI in Pucheng 
and as /n/ before the nasal ending and HI elsewhere in Wuyi. Moreover, 
the /1/s in both places occur in upper tone categories. Pucheng and Wuyi 
where the Duan initial is realized as HI are located on the south and north 
ends of the area where it is realized as fldj. 

(3) Realization of the Palatal Affricates as the Zero Initials. 

Palatal affricates are realized as zero initials in some dialects in 
midwestern Zhejiang, as in 'till the land'^f @ ly*\ cdi£ /, 'swollen* 9i^ 
/ c y9f)l, 'bracken' $$j}|- / 5 loni/, 'ditch* #M| (for irrigation of paddy field), 
(<Jiyun> zhu run qie ^?t\iu) /yan*/ in the Songyang dialect. The 
realization of preglottalized palatal affricates as zero initials is quite 
common in the Kam-Tai languages. Consider the following examples in the 
Buyi language: 

N be/live' x wink' ; beat /fish out 1 

2d*bp - d^ok 

*jfeP 2 P 

Ijtp * ok 

J*P jok 

Zhengzhang (forthcoming) first noticed this and held that the realization of 
some Middle Chinese obstruents as zero initials is the outgrowth of 
preglottalized palatal obstruents (**d - >*j >i). The phonetic correlates 
remain unexplained. It may have to do with the loosening of the closure of 
the backward tongue caused by the glottal constriction. Those occurrences 
of zero initials distributes over Longquan, Qingyuan, Jingning, Taishun, 
Yunhe, Lishui, Qingtian, Jinyun, Yongkang, Dongyang, Suichang, 
Songyang, Longyou, Lanxi, Jiangshan, Quxian, Changshan and Kaihua in 

Xingyi Bajie 

*o> u 

Huishui Changan 

1 jiu 

An long Bakan 

1 jiu 

Li po Chaoyang 



Zhejiang (the data is due to Zhengzhang). The distribution of this kind of 
zero initials and that of implosives are coterminous. Moreover, Jiande and 
Chun'an have the same phenomenon, showing the relationship between the 
dialects in this area and Wu dialects. 

We can see that with the exception of northern Zhejiang and the 
western part of southern Jiangsu various places of Wu dialects have 
implosives and their traces. Thus, Bang and Duan initials were realized as 
implosives in proto-Wu dialects and the existence of /* i g-/, /* d$,-/ in 
the same dialect group can be also postulated from the occurrence of zero 
initials. In southeast Asia implosives mainly distribute over the realm of 
ancient Yue peoples, as in the Kam-Tai language, Vietnamese, the 
Wenchang dialect in Hainan island and some other dialects. Implosives in 
Wu may well be the substratum left by the Yue language. 
2.1.2 The Distinction of Vowel Length in Wu Dialects. 

The distinction of vowel length is an important feature of southeast 
Asian languages such as the Kam-Tai language and Vietnamese. There is a 
contrast in vowel length in some rimes of Cantonese. In general, Wu 
dialects have no contrast in vowel length and hence the dialectal literature 
devotes scanty attention to the discussion of vowel length. However, a 
handful of dialects that the present author surveyed do show a contrast in 
vowel length, most of which have no phonemic values. The point is clear j 
when diphthongs in the Wenzhou dialect and Pekingese are compared. A | 
comparison of / c kai/ 'return' in the Wenzhou dialect and / c kai/ 'should* in 
Pekingese reveals that there is virtually no qualitative segmental and tonal 
difference; what is an apparent difference is the length of the vowel /a/. 
^kai/ in Pekingese is close to <cai> 'should' in Vietnamese, whereas /<kai/ 
in the Wenzhou dialect approximates <cay> 'spicy hot' in Vietnamese 
Likewise, / c kau/ 'tall' in Pekingese is near <cao> 'talP in Vietnamese, while] 
/ c kau/ 'hook' in the Wenzhou dialect is approaching <cau> 'betel nut'. In the] 


Wenzhou sound system a vowel with no ending will be realized as a long 
vowel and as short vowel when followed by endings. Although Wenzhou 
speakers render Pekingese vowels accurately, they still cannot cover up 
their accent, one of the reasons being that the main vowel in the diphthong 
is shortened. /9u/ and /^y/ in the Suzhou dialect has an iambic prosodic 
pattern (i.e., both, short + long and weak + strong). Therefore we might just 
as well regard the ^ as a monophthong as a long vowel and the ^ in a 
diphthong as a short vowel. Medials /i, u, y/ in most Wu dialects are linked 
up with initials, and cannot be lengthened. In actuality, they are short /i, u, 
y/. Thus, /i, u, y/ fall into two types: the long ones function as main 
vowels and the short ones as medials. Most intriguing are the cases in 
Yongkang and Qingtian dialects where medials show a contrast in length, 
/i, u, y/ and /1, u, y/, as in: yuany\i\ / c yd/, yu1tffyhl\ jieHfciB/, 
ji A^fch /; ke <i% /k'ua /, qu # /k'uV/ in Yongkang (Yuan et al 1960) and 
gan g^/ c kua/, guan sg /^kfia/; xiao '^ Ipio / , xiao-^f Ijtiol. Vowels /a/ 
and hi after a long medial are virtually neutralized as M/ in Qingtian. 

There is a whole set of contrasts in vowel length in Kam-Tai 
languages; Cantonese shows only partial contrast; Wu dialects have long and 
short vowels which, however, are not in contrast; Mandarin has no 
distinction in vowel length. In the geographical distribution of linguistic 
typologies Wu dialects are located on the intermediate zone between the 
type of contrast of vowel length and the type of no contrast of vowel 
length, most possibly the result of the mixture of the Han language and the 
proto-Yue language. 
2.1.3 The Unaspirated Voiceless Stops in Wu Dialects Being Stiff and 

Crisp with Strong Explosion. 

This articulatory trait sets them apart from other major dialects. The 
Wu speakers produce a rather "stiff" version of Pekingese unaspirated stops. 
This is one of the main reasons that they immediately give themselves away 


as Wu speakers. It is quite likely that this characteristic of stops has to do 

with the preservation of voiced obstruents. 

2.1.4 The Unroundedness of the High Back Vowel. 

The further south we go the less round the vowel /u/ is. In southern 
Zhejiang dialects the vowel /u/ has become unrounded. But here it is 
phonemicized as /u/, but not M/, since, as far as the lip position is 
concerned, /iu/ is open and yet this sound is close and its timbre is closer 
to /u/ than /ux /. /u/ has developed into li> I in the Wenzhou area, a 
vocalized version of /v/. The unrounded /u/ not only appears in Wu 
dialects, but in Gan, Hakka and Min dialects as well. This kind of 
convergence may well represent some kind of relationship. The 
unroundedness of their /u/ is also one of the main reasons that Wu speakers 
cannot easily shed their own accent in learning Pekingese. 





edited by 
William S-Y. Wang